Why Assembly
We measure the correction, not the completion.
We measure the correction, not the completion.
Completion rates are the easiest number to report and the least useful one to act on. Assembly is built around what happens after the agent answers.
Outcome over completion — a run that finished is not a run that worked.
Vendor-agnostic — one record across every agent platform you run, not five dashboards.
Scope-aware by design — we record what an agent was permitted to read, and record no more than that.
Published methodology — including the corrections we have had to issue ourselves.
0%0% Median silent correction rate00 Organisations in the Index00 Record across all vendors
In their words
Fenwick Council
The governance conversation was harder than the deployment. It should be.
Workplace Technology Lead, Fenwick Council
Ordnance Trust
Finance did not want an opinion on AI. They wanted the arithmetic.
Head of Digital, Ordnance Trust
Quantic Systems
We had eleven agents doing overlapping work and no way to see it.
Support Operations Lead, Quantic Systems
Get started
Judge us on the research.
Judge us on the research.
Every benchmark we publish ships with its methodology and its corrections.
Browse the researchGet started
Judge us on the research.
Judge us on the research.
Every benchmark we publish ships with its methodology and its corrections.